RenaSciQuestion network · The Renaissance of Science E6Fractal Curation

E6HypertextFurther topic

Fractal Curation

Scholarly Works in the Age of Hyperlinks

?If a scholarly work is no longer a paper, what should it look like?

Topic page
~876 words · 8 screens
Deep read
~7,623 words · 85 sources
Questions
14 out · 16 in
Start with the first question ↓Go straight to the deep read ↗
  1. 01What do readers need?
  2. 02Who was hypertext invented for?
  3. 03Why didn’t hypertext change the paper?
  4. 04How many layers can a work have?
  5. 05How can every entrance be a starting point?
  6. 06As links multiply, who vouches for them?
  7. 07Does mapping knowledge count as scholarship?
  8. 08Does this network practise what it preaches?

◇The network re-centres on “Fractal Curation”. Every screen leaves a few follow-up questions; open one to walk into another topic.

01/08Map

What do readers need?

Readers need maps, not more photographs

A paper is like a photograph: carefully composed, cleanly framed, frozen at the instant the shutter clicked. A research field is more like a city whose streets change every day; the open index OpenAlex already lists more than 477 million scholarly works. Faced with a city like that, one more photograph does little good. Newcomers and old residents alike need a map: where to come in, what to see first, which road leads where.

Follow-up questions
Read deeperWhat do readers need?↘
Asked from elsewhereE2Flood

02/08Trails

Who was hypertext invented for?

Trail blazers and associative trails

In 1588 Agostino Ramelli designed a “book wheel”: turn the wheel, and you could move back and forth between several books. In 1945 Vannevar Bush, who had directed America’s wartime research, imagined a desk called the memex that would string documents together with associative trails – trails you could even copy for a friend. He predicted a new profession: trail blazers. In 1989 Tim Berners-Lee proposed the World Wide Web at CERN; the first website in the United States served physics preprints at the Stanford Linear Accelerator Center. From the very beginning, hypertext was made for researchers.

Follow-up questions
Read deeperWho was hypertext invented for?↘
Asked from elsewhereE1LettersE2Flood

Ramelli’s “book wheel”: turn the wheel to move between many open books. Plate 188 of Le diverse et artificiose machine, 1588. Wellcome Collection · Public domain

03/08Paper shadow

Why didn’t hypertext change the paper?

It changed everything except the paper

Decades later, the paper still looks much like paper; it has simply become a PDF. Academic credit is tied to texts that can be cited and never change. Links themselves are unreliable, too: a 2014 study of more than 3.5 million papers in science, technology and medicine found that in one paper in five, the cited web pages could no longer be seen as they had been. More fundamentally, people always judge a new technology by the concepts of the old: the car was seen as a carriage without a horse, the web page as a superior page of a book. Then social platforms pressed everything into the feed. On the Chinese-language internet, personal websites all but vanished, and scholars’ home pages and blogs withered with them.

Follow-up questions
Read deeperWhy didn’t hypertext change the paper?↘
Asked from elsewhereE1New bodyE2Digital paper

Interior of the Bodleian Library, Oxford. Line engraving by Loggan. Wellcome Collection · Public domain

04/08Seven layers

How many layers can a work have?

Seven layers, each in its place

A work can have seven layers: core argument, traceable evidence, process record, multimodal explanation, governance transparency, AI interface and collaborative evolution. Layering does not mean chopping a book into pieces. Some things cannot be compressed: a novel cannot be summarized, and an argument has its own rhythm. So there has to be a room for immersion, where those who want to read to the end can do so. Let the scrollers scroll and the readers read: both are legitimate ways of reading.

Follow-up questions
Read deeperHow many layers can a work have?↘
Asked from elsewhereMNew bodyE1Books

05/08Fractal

How can every entrance be a starting point?

One question leads to another

News writing uses the inverted pyramid, putting the most important thing first. Stretchtext lets readers unfold the details layer by layer. Mathematicians break large proofs into clickable dependency graphs that show at a glance which steps have been proved and which have been formalized. We bring these methods to curation: one page per topic, making clear what it is asking; the follow-up questions on the page lead to other topics, and those who want to go further can dive down into the deep read, the data and the process. The topic page is the gate, the deep read the threshold. The further in you go, the more is asked of you.

Follow-up questions
Read deeperHow can every entrance be a starting point?↘
Asked from elsewhereMNew body

06/08Witness

As links multiply, who vouches for them?

Trust comes from layer upon layer of texts

For a network to be trustworthy, every path has to be traceable: versions frozen and archived, citations pinned to the paragraph and the version, changes entered in a log, AI’s involvement in plain sight. Mathematicians have already demonstrated the strongest form of witnessing: the Equational Theories Project had Lean check every one of the more than 22 million implications among 4,694 algebraic laws. But machines can witness that a proof is right; they cannot yet witness that it is understood. A proof that nobody can explain is not complete, even if it has been verified.

Follow-up questions
Read deeperShould works be machine-friendly?↘
Asked from elsewhereMStandardsE3TheatreE4SeriousE5Maths

An Experiment on a Bird in the Air Pump. Mezzotint by Valentine Green after Joseph Wright, 1769. Wellcome Collection · Public domain

07/08Curator

Does mapping knowledge count as scholarship?

The curator: a new scholarly role

Arranging a network like this calls for a new role. The word “curate” goes back to the Latin for “to take care of”: selecting, arranging, explaining – and then maintaining the work year after year, repairing dead links, answering critics. Bush foresaw “trail blazers”; the two editors of Distill argued that “distillers”, who make research clear, are as indispensable as theorists and experimentalists. This work deserves academic credit. In the age of AI the scholar is more like a wine taster: machines can analyse a wine’s composition, but they cannot replace the taster, because only people love wine.

Follow-up questions
Read deeperDoes mapping knowledge count as scholarship?↘
Asked from elsewhereE3ThresholdE5Humanities

The cabinet of natural and artificial curiosities at the Bibliothèque Sainte-Geneviève, Paris (detail). Engraving. Wellcome Collection · Public domain

08/08Experiment

Does this network practise what it preaches?

This set of topics is an experiment

Networks have their costs, too: readers get lost, links rot, maintenance costs money, and a work may never be finished. So there has to be a clear starting point, layers and a map, and versions have to be frozen and archived at regular intervals. The set of topics you are reading is itself an experiment: the manifesto is the starting point, and the seven further topics each ask their own question while tugging at one another. It will keep being revised, and you are welcome to point out where you hit a dead end.

Follow-up questions
Read deeperWill readers get lost in the network?↘
Asked from elsewhereMNew body

E6The discussion goes on

The discussion goes on

?If a scholarly work is no longer a paper, what should it look like?

Hypertext was originally invented for scientists. In 1945 Vannevar Bush imagined stringing documents together with “associative trails”; in 1989 Tim Berners-Lee proposed the World Wide Web for the physicists at CERN. Eighty years on, the basic unit of scholarly output is still the paper, each one an imitation of the printed page. This essay traces that history, explains why hypertext failed to change the paper, and proposes layered works and fractal networks: every act of curation is a question network. We also discuss the precedents, what this means for machine readers, how it can earn trust on the web, the new role of the curator, and the risks. And this set of topics is itself an example.

Deep readFractal Curation: Scholarly Works in the Age of Hyperlinks~7,623 words · 11 questions · 85 sources↗
  1. 1What do readers need?
  2. 2Who was hypertext invented for?
  3. 3Why didn’t hypertext change the paper?
  4. 4How many layers can a work have?
  5. 5How can every entrance be a starting point?
  6. 6Has anyone done this already?
  7. 7Should works be machine-friendly?
  8. 8As links multiply, who vouches for them?
  9. 9Does mapping knowledge count as scholarship?
  10. 10Will readers get lost in the network?
  11. 11Does this network practise what it preaches?

→Questions from here

The follow-up questions left on this page, grouped by the topic they lead to.

←Questions from elsewhere

Questions asked in other topics that lead here. Open one to return to the screen where it was asked.